Last year we presented a spreadsheet of what the National Nuclear Security Administration (NNSA) was estimating would be the total costs of pit production at Los Alamos National Laboratory (LANL) and the Savannah River Site (SRS) through the 2020s in slides 23-31 here. Per-pit costs were estimated by Congressional Budget Office (CBO) for SRS (slide 28) and we offered alternative methods of calculating per pit costs at LANL.

There have been some developments since then. Estimated capital costs at SRS and LANL have both increased. Budget details beyond fiscal year 2022 (FY22) are however lacking. The Biden Administration's May 30 budget request (Administration budget for warhead cores ("pits") jumps again, as Biden team struggles with Trump's rushed "double-factory" plan, June 1) didn't include a five-year spending plan, and the biggest capital projects at each site had no year-by-year schedules. There is also no up-to-date staffing plan for SRS available to the public at present.

It is still possible to calculate ballpark pit costs under a range of reasonable assumptions, and we have done that.

LANL's main capital project for pits -- the Los Alamos Plutonium Pit Production Project (LAP4) -- is supposed to be done in FY28. If we assume that 121 pits are produced by FY28 (1, 10, 20, 30, 30, 30 in years 2023 to 2028, respectively), LANL's expenses for pits over that period will be approximately:

• $14.2 billion (B), if we include "Plutonium Modernization" expenses over the FY19-FY28 period and prior-year costs of the Chemistry and Metallurgy Research Replacement (CMRR) and the TA-55 Reinvement Project Phase III (TRP III), and omit all other expenses. This leads to a cost of $117 M/pit.
  
• $9.2 B, if we omit all sunk costs and include costs for FY23-FY28 only, or $76 M/pit.

We use the high-end estimate for LAP4 ($3.9 B) rather than the mean or the low-end estimate ($2.7 B). We cannot immediately think of a case when the high-end estimate was not the most accurate of those three.

This is really too short a time-period to be representative. So if instead we assume 211 pits are produced by 2033, and omitting all costs incurred up to and including this year, LANL's expenses by then will be:

• $15.2 B, leading to $72 M/pit, assuming annual expenses are $1.2 B/yr for the final 5 years ($1 B for 4,000 staff and $0.2 B for infrastructure each year).
  

We are making a much more optimistic assumption about the stability of LANL production than NNSA. That said, the farther one goes into the 2030s the more likely additional large-scale expense to support pit production will be needed at LANL.

With that in mind, considering a longer timeframe, through 2039, with 391 pits built, LANL's expenses would be:

• $22.4 B, or $57 M/pit, omitting all costs incurred up to and including this year and including only $1.2 B per year after that (i.e. no major new infrastructure)
  
• $32.4 B,  or $83 M/pit, including $10 B for new nuclear facilities.
  

These costs do not include costs ("Enterprise Support") incurred at other sites to support LANL pit production, which from last year's 5-year spending plan we might guess to level out at about $200 million per year, or almost $7 M per LANL pit.

These costs also do not include the "seven" construction line items already built to support pit production, mentioned but not identified in the 2019 LANL Pit Production Plan. Nor do they include future line item construction identified in the FY21 Stockpile Stewardship and Management Plan (SSMP) but not included in the FY22 budget request, e.g. Sigma Replacement (estimated for FY26-37), a Radiography/Assembly Complex Replacement (RACR, estimated for FY32-41. Sigma Replacement alone would a billion-dollar project. As far as we know, Sigma does not meet federal seismic codes. Neither do other LANL buildings needed for pit production.)

These costs also do not include "smaller" construction projects supporting pit production which are not congressional line items, such as multilevel parking garages, other parking areas, offices, cafeterias, security headquarters, and training centers. We have good visibility for most of these through FY22 but they appear without warning each year. These costs may add up significantly.

Except as noted, line item projects to augment or replace Building PF-4 or to extend its life are not included in these costs. These will be necessary sooner or later if production continues, or under contingencies. These will cost multiple billions and take many years to build, if indeed they can be built at all at LANL.

Nor are any pro-rata costs for site-wide and corridor-wide improvements included, such as roads, new electric transmission lines, and other utilities.

We do not know how many pits per year LANL can produce. No one does. LANL Director Thom Mason has recently said that it will be difficult to reach 30 ppy by 2026. We have used an average of 30 ppy from 2026 on, which we believe is what NNSA is now saying is its maximum steady-state production at LANL.

In sum, $50 M/pit appears to be a conservative number for LANL.

The marginal pit cost of $6 M quoted by CBO and in our slide 28 for SRS has not much changed so far, as it is not affected by increased capital costs.

Using the best public data, it appears LANL pits will cost roughly an order of magnitude more than SRS pits and at least 5 times the total non-pit costs of the W87-1 warhead as estimated by CBO and independently by us last year. W87-1 warheads made with LANL pits would cost about 4 times what W87-1 pits made with SRS pits would cost, again using the best available public data.